Cloud-native EV charging platform
Shell Recharge runs a public EV charging network across the UK, EU and US, with hundreds of thousands of chargepoints. I led the move of its entire backend, including the Charge Point Management System (CPMS), from co-located VMware to a cloud-native architecture on AWS, and built the SRE practice that runs it.
- Role
- Senior Technical Lead, Platform and SRE, then Solutions Architect
- Team
- Cross-functional team of 20; direct team of 8
- Scope
- UK, EU and US · hundreds of thousands of chargepoints · 100+ microservices
The challenge
The backend, including the CPMS, ran on ageing co-located Cisco UCS and VMware infrastructure. Availability was poor, releases were slow and risky, and costs were rising as the network grew across Europe and the US.
What I did
- Led the migration to containerised workloads on AWS EKS, with GitOps delivery through ArgoCD and an Istio service mesh.
- Set up the SRE function: SLIs and SLOs for critical services, blameless post-mortems and production-readiness checks.
- Rolled out Datadog and OpenTelemetry observability built around golden signals, with monitoring as code.
- Designed a predictive maintenance pipeline for chargepoints on AWS IoT Core and SageMaker.
- Moved the payments database to Amazon Aurora PostgreSQL.
Outcome
- 69% → 99.8%platform uptime within six months
- 99.99%reliability after the cloud-native migration
- −20%operational cost
- −25%faulty chargepoints with predictive maintenance
Stack
- AWS EKS
- ArgoCD
- Istio
- Cilium
- KEDA
- Terraform
- Datadog
- OpenTelemetry
- AWS IoT Core
- SageMaker
- Aurora PostgreSQL